Lecture recording during the workshop Advanced Neural Data Analysis (ANDA)

Lecture recording during the workshop Advanced Neural Data Analysis (ANDA) 2019.

Philosophical theories of ethics, mental privacy, ethical aspects of animal experiments, ethical aspects of clinical neuroscience and patient research, good scientific practice, data protection and computer security, neurolaw, ethics committees.

This course is intended as a refreshment of mathematical tools of analysis, linear algebra and statistics which will be necessary for the CNS students in the first year. Students will acquire broadmathematical knowledge of functions in one resp. several real variables, in linear algebra, in differential equations, in probability theory and statistics, as needed for Computational Neuroscience. Basic mathematical skills for the analysis and approximation of functions, solutions of differential equations and signals, for solving linear systems and systems of ordinary differential equations will be refreshed.
Participants will learn to apply mathematical foundations to the modeling and analysis of neural data and to use basic mathematical techniques to problems in Computational Neuroscience with guided assistance.

This course is intended as bridge for students without physiology training, enrolling in Computational Neuroscience. The aim is to provide the basics in neurophysiology.

Course description:
Participants learn basic concepts, their theoretical foundation, and the most common algorithms used in machine learning and artificial intelligence. After completing the module, participants should understand strengths and limitations of the different paradigms, should be able to correctly and successfully apply methods and algorithms to real world problems, should be aware of performance criteria, and should be able to critically evaluate results obtained with those methods. Participants should also be able to modify algorithms to new tasks at hand as well as to develop new algorithms according to the paradigms presented in this course.

Contents include:
Artificial neural networks: Connectionist neurons, the multilayer perceptron, radial basis function networks, learning by empirical risk minimization, gradient-based optimization, overfitting and underfitting. Learning theories and support vector machines: statistical learning, learning by structural risk minimization.
Probabilistic methods: Bayesian inference and neural networks, generative models. Projections methods. Principal Component Analysis, Independent Component Analysis and blind source separation. Stochastic optimization. Clustering and embedding.

Students will gain knowledge about the most important methods for experimental acquisition of neural data and the respective analytical methods, they will learn about the different fields of application, the advantages and disadvantages of the different methods and will become familiar with the respective raw data. They will be enabled to choose the most appropriate analysis method and apply them to experimental data.
Contents include:
Acquisition of neural data (1st semester): large scale signals (fMRI, EEG, MEG etc) and cellular signals, hands-on experience with neural data acquisition techniques.
Analysis of neural data (2nd semester): firing rates, spike statistics, spike statistics and the neural code, neural encoding, neural decoding, discrimination and population decoding, information theory, statistical analysis of EEG) data, spatial filters, classification, adaptive classifiers.

Participants should learn the basic concepts and most important topics in the Cognitive Neurosciences. In addition, they should know the state-of-the-art models in these domains and their theoretical foundations. After completing the module, participants should understand strengths and limitations of the different modeling approaches (e.g. bottom-up versus top-down), should be able to understand the rationale behind models and their implementation, and should be aware of performance criteria and critical statistical tests. Participants should also be able to modify models of cognitive processes as well as to apply existing models to novel experimental paradigms, situations or data.
Contents include:
Auditory and visual system, natural image statistics and sensory processing, motor system, psychology and neuroscience of attention, memory systems, executive control, decision making, science of free will and consciousness. Data modeling and essential statistics, psychometric methods, signal detection theory, models of visual processing, models of visual attention, models of executive function. Signal processing, sensory and cognitive modeling using Python.

Participants should learn basic concepts, their theoretical foundation, and the most common models used in computational neuroscience. The module also provides the relevant basic neurobiological knowledge and the relevant theoretical approaches as well as the findings resulting form these approaches so far. After completing the module, participants should understand strengths and limitations of the different models. Participating students will learn to appropriately choose the theoretical methods for modeling neural systems. They will learn how to apply these methods while taking into account the neurobiological findings, and they should be able to critically evaluate results obtained. Participants should also be able to adapt models to new problems as well as to develop new models of neural systems.
Contents include:
Hodgkin-Huxley model, Channel models, Synapse models, Single-compartment neuron models, Models of dendrites and axons, Models of synaptic plasticity and learning, Network models, Phase-space analysis of neuron and network models (linear stability analysis, phase portraits, bifurcation theory).

Participants should learn basic concepts, their theoretical foundation, and the most common models of stochastic evolution equations on Hilbert spaces with a view towards its applications to the modelling, analysis and numerical approximation of spatially extended neurons and neural systems subject to noise. Participants will learn basic techniques to analyze global properties of neural systems both qualitatively and quantitatively. Participants will also learn basic simulation techniques for stochastic neural systems and how to evaluate simulation output. Participants should also be able to adapt models to new problems as well as to develop new models of neural systems.
Contents include:
Gaussian measures on Hilbert spaces, stochastic integration on Hilbert spaces, semilinear stochastic evolution equations, stochastic reaction diffusion systems, continuum limits of neural networks.

Participants should learn basic concepts, their theoretical foundation, and the most common models of stochastic processes used in computational neuroscience to model noisy neural systems. Participants will learn basic techniques to analyze the stochastic behavior of singles neurons and neural systems both qualitatively and quantitatively. Participants will also learn basic simulation techniques for stochastic neural systems and how to evaluate simulation output. Participants should also be able to adapt models to new problems as well as to develop new models of neural systems.
Contents include:
Brownian motion and stochastic calculus, stochastic models for single neurons (stochastic Hodgkin-Huxley model, stochastic integrate-andfire models, random oscillators), coupled neurons with noise, synchronization, stochastic stability, stochastic neural fields, travelling waves.

The computational neuroscience group at the Ruhr University Bochum studies the cognitive and neural mechanisms underlying episodic memory and spatial navigation. The group employs a wide range of computational methods ranging from (spiking) neural networks, abstract cognitive models to machine learning algorithms such as unsupervised and reinforcement learning.

Practical course on neuronal network modeling and data analysis. Problems selected from the fields of neuronal coding, neuronal decoding, data analysis and small neuronal networks will be treated in small projects. The focus lies on the development of simple, yet viable models and the performance of computer simulations required for obtaining significant results.

Learn to write your own computer programs to analyse data and simulate neuronal systems. In the first half of the lectures and practical exercises, you will achieve the basic skills to write computer programs which perform simple calculations, and we will advise you how to break down a more complex problem into simple tasks a computer can perform. In the second half of the course, you will apply your acquired skills to analyse neural signals (mean and variance, estimation of firing rates, reverse correlation, ROC analysis, etc.), and simulate single neurons or synapses (integrate-and-fire neuron, Hodgkin-Huxley neuron).

Introduction to fundamental concepts in Computational Neuroscience. In the first term, we will study basic encoding and decoding schemes, analysis of neural signals, and the dynamics of single neurons. In the second term, we will focus on synapses and neural networks, and study emergent phenomena such as computation and classification, learning and memory, pattern formation, and synchronization.

Follow-up to course "Computational Neuroscience I" held in Winter Semester: Here we will focus on synapses and neural networks, and study emergent phenomena such as computation and classification, learning and memory, pattern formation, and synchronization.

This advanced course aims at providing deeper insights in state-of-the-art questions in neuroscience, analysis approaches and how to formalize questions to neuronal data so they can be answered quantitatively. The course addresse ecellent master studetns and PhD students interested in data analytics and in getting hands-on experience in the analysis of electrophysiological data (multiple-parallel spike trains and local field potentials). In the first week of the course, international experts will give lectures on statistical data analysis and data mining methods with accompanied exercises. In the second week the participants will pursue their own data analysis projects on a common data set.

This module will equip the student with basic skills of scientific programming with PYTHON and provide the student with hands-on experience in the statistical analysis of experimental neurophysioloigcal data sets and the adequate presentation of results. No previous programming skills are required.

This lecture was recorded during the Workshop "Advanced Neural Data Analysis Course" (ANDA) 2019.

Advanced introduction to the implementation of cognitive models; levels of description: computational, algorithmic, and implementational models; reading, understanding, and implementing recent publications involving cognitive models, e.g. human information processing, decoding of physiological signals, artificial cognitive systems, machine learning in psychology, motor control.

This multi-semester course focuses on emerging interdisciplinary
perspectives on ‘Systems Science and Engineering for Intelligence’. We focus on natural systems, human and computer vision, bio-inspired vision system designs, and systems theory required for modeling, analysis, simulation and validation of cognitive vision systems. The emphasis is on abstractions, modeling, and rigorous statistical approaches to performance evaluation. Connections are also made between engineering designs and architectural
designs in natural systems. The course draws upon years of systems engineering – involving systems modeling, analysis, and validation of computer vision systems and explores links them to latest viewpoints from Machine Learning, Artificial Intelligence, and Brain Sciences. The core emphasis in the course is that there is a natural correspondence between (Application Contexts, Tasks, Performance Requirements) and (Hardware and Software Programs and their Parameters). Paradigms popular for systems design – model-based systems engineering vs data-driven machine learning
will be contrasted. The course material is largely based on dissertations, publications, and online video material.
The objective of the course is to teach foundations in systems thinking that can be applied to design, analysis, and validation of intelligent systems. Through case-studies in computer vision the students learn systems modeling, simulation and optimization of intelligent systems.

Supervised, unsupervised and semi-supervised learning, reinforcement
learning, Bayesian learning, Energy minimization and optimization.

We will discuss the principles guiding the formation of sensory maps and receptive fields during circuit development. The students will examine how different mechanisms including: emergence of diverse single neuron properties and activity-dependent synaptic plasticity interact during development to give rise to functional circuits. The students will have the opportunity to analyze data from visual cortex and build their own models of the assembly and tuning of developing neuronal circuits.

This module provides an introduction to modern theoretical neuroscience with an attempt to cover all relevant spatial scales (from molecules to brain areas) as well as temporal scales (sub- millisecond to evolutionary times scales).

This is an introductory-level seminar series on the psychology and biology of visual perception. Topics include illusions, retina, contours, colour, motion, Gestalt psychology, faces and objects, attention and consciousness, computer vision, eye movements, neuropsychology and art. Each week 2-3 students present one paper each for 20 minutes. After each presentation there is a question and answer and discussion section. Each student can also receive a 30-min tutorial with me a few days before their presentation. Participants receiving five or more CP must also complete an essay on the same topic as their presentation.

Masters projects for motivated students seeking a career in research are available in Roland Fleming's lab.  Research in the lab focusses on mid- and high-level vision, including the perception of 3D shape (e.g. shape-from-shading) and the physical properties of objects and materials (e.g. viscosity, stiffness). We make extensive use of photorealistic computer graphics and simulations. The project would combine psychophysics with modeling (e.g. deep learning).

Students will learn to estimate active information storage, transfer and modifcation using the IDTxl toolbox. datasets will be electrophysiological recordings from non-human primates and MEG data from patients with Autism.

An introductory meeting will take place on Wednesday, 24th of October at 12:15 at the MPI for Dynamics and Self-Organization, room 0.77
The block seminar will take place in the week 18.-22. February 2019
Programming skills are required. The number of participants is limited to 15.
Inspired by the architecture of the human brain, artificial neural networks have been developed in the past decades to process and solve various types of tasks. To understand processing in such networks, a natural framework is provided by information theory. The seminar will focus on information theory to study the design, dynamics, and function of such networks.
The seminar is organized as a block seminar, in the spirit of summer schools or code jams. The students choose a topic to work on in small groups, preparing the background theory, implementing the main results, and ideally obtaining novel insights based on their own research projects within the block seminar.

Collective dynamics; subsampling; information theory; synaptic plasticity; principles of local learning.
Short research projects or collaborations comprising any of these topics are defined together with the student.

Introduction to Neurophysics, Modeling and Methods, Nonlinear Dynamics, Statistical Physics, Neurobiology, Neural Networks. This lecture course offers an introduction to advanced modeling strategies for biological neural networks. After a short introduction to the biophysics of single cells and an overview of their basic firing patterns, we explain fundamental properties of networks models of neurons, starting from simple uniform connectivity and progressing to spatially extended and to arbitrarily complex interaction networks. These network models explain and predict key dynamical aspects of neural circuits, including irregular activity of cortical dynamics, feature selectivity, self-organization of neural maps, and the coordination of precisely timed spikes across networks. The summer term course has its focus on neural field models.

The  small private online course (SPOC) "Network Neuroscience" provides a broad overview on recent research regarding Network Neuroscience with the following topics presented by members of the SFB 936:

Statistical theory of equilibrium states (e.g. collective phenomena, phase transitions, renormalization group theory) and nonequilibrium processes (e.g. kinetic theory, stochastic processes).

Students learn the principles, on which the modeling of complex systems is based, as well as the mathematical methods that are employed in deriving and solving such models.

The neural networks of the brain form one of the most complex systems we know. Many qualitative features of the emerging collective phenomena, such as correlated activity, stability, response to inputs, chaotic and regular behavior, can, however, be understood in simple models that are accessible to a treatment in statistical mechanics. This course presents the fundamentals behind contemporary developments in neural network theory that are based on methods from statistical mechanics of classical systems with a large number of interacting degrees of freedom.
The focus on classical systems allows us to introduce the standard language and tools employed in statistical field theory in a simple and didactic form moments, cumulants, generating function[al]s, Wick's theorem, linked cluster theorem, perturbation theory, Feynman diagrams, mean-field approximation, loopwise expansion). We will explain and derive these concepts as far as they are needed in the context of neural networks. This
first part will be familiar to students with some knowledge in (quantum) field theory and statistical physics.

Models of neurons, synapses and networks; concepts of neuronal coding and cortical information processing; plasticity and learning. Data analysis and visualization by self-written programs; Usage of scientific programming languages (Matlab and Python), also for documenting the analyses; hypothesis tests by numerically generated modified data ('surrogate data'); Simulation of neuronal circuits.

• Markov models of network activity
• Mean-field theory of binary/spiking neuronal networks
• Fokker-Planck theory of spiking neurons
• Linear response theory of correlations
• Correlation structure of activity in recurrent neuronal networks

In this course, I plan to treat the dynamics of populations of neurons and of synaptic learning.

Special mathematical and statistical tools are required to model and analyze biologically plausible networks of spiking neurons. The course offers a step-­by-­step introduction to stochastic dynamical systems up to and including current mean-­field approaches to recurrent networks of spiking neurons. Stochastic variables, stochastic processes, interval distributions, autocorrelation and power spectrum, Wiener process and white noise, LIF neurons with Poisson input, diffusion limit, Fokker-­Planck equation of VIF neuron, mean-­field theory of recurrently connected populations, application to decision making and confidence. In the practical part, students simulate and analyze recurrent networks of spiking neurons on the basis of Matlab and Neuron.  

Based on Chapters 5-­6 and Chapters 1-­4 of Dayan & Abbott. Electrochemical equilibrium and Nernst Equation, equivalent circuits for single-­compartment model, leaky integrate-­and-­fire model, Hodgkin-­Huxley and Connor-­Stevens models of action potential, cable equation and neuron morphology, characterizing neuronal responses with tuning curves and receptive fields, signal-­ detection theory and psychometric function, comparison of neuronal and behavioural responses with neurometric function, population coding, statistically efficient decoding with maximum likelihood and maximum a posteriori likelihood, introduction to Shannon information, application of Shannon information to neural responses.

Computational neuroscience is a highly interdisciplinary field ranging from mathematics, physics and engineering to biology, medicine and psychology.

Interdisciplinary collaborations have resulted in many groundbreaking innovations both in the research and application.

The basis for successful collaborations is the ability to communicate across disciplines: What projects are the others working on? Which techniques and methods are they using? How is data collected, used and stored?

In the MidsummerBrains colloquium, experts from the SMARTSTART faculty describe their view on computational neuroscience in theory and application, and share experiences they had with interdisciplinary projects.

This is a recording of the lecture "There and back again: Computational neuroscience and field biology", held by Jan Benda, head of the department of neuoethology at the University of Tübingen (Dep. of Neuroethology).

Jan Benda talked about his point of view on computational neuroscience, the tricky life beyond lab conditions and quite explicitly talking fish. 

Computational neuroscience is a highly interdisciplinary field ranging from mathematics, physics and engineering to biology, medicine and psychology.

Interdisciplinary collaborations have resulted in many groundbreaking innovations both in the research and application.

The basis for successful collaborations is the ability to communicate across disciplines: What projects are the others working on? Which techniques and methods are they using? How is data collected, used and stored?

In the MidsummerBrains colloquium, experts from the SMARTSTART faculty describe their view on computational neuroscience in theory and application, and share experiences they had with interdisciplinary projects.

This is a recording of the lecture "My View on Computational Neuroscience", held by Tatjana Tchumatchenko, head of the theory of neural dynamics group at the MPI for Brain Research in Frankfurt (tchumatchenko.de/).

She talked about her point of view on computational neuroscience as a physicist and the winded path from experiment to insight and publication.

Computational neuroscience is a highly interdisciplinary field ranging from mathematics, physics and engineering to biology, medicine and psychology.

Interdisciplinary collaborations have resulted in many groundbreaking innovations both in the research and application.

The basis for successful collaborations is the ability to communicate across disciplines: What projects are the others working on? Which techniques and methods are they using? How is data collected, used and stored?

In the MidsummerBrains colloquium, experts from the SMARTSTART faculty describe their view on computational neuroscience in theory and application, and share experiences they had with interdisciplinary projects.

This is a recording of the lecture "Quantifying nonlinear representations in the visual systems", held by Fabian Sinz, group leader of the neuronal intelligence lab in Tübingen as part of the CyberValley initiative (sinzlab.org/).

He is talking about the difficulties of nonlinear representations and what black boxes mean for machine learning.

Computational neuroscience is a highly interdisciplinary field ranging from mathematics, physics and engineering to biology, medicine and psychology.

Interdisciplinary collaborations have resulted in many groundbreaking innovations both in the research and application.

The basis for successful collaborations is the ability to communicate across disciplines: What projects are the others working on? Which techniques and methods are they using? How is data collected, used and stored?

In the MidsummerBrains colloquium, experts from the SMARTSTART faculty describe their view on computational neuroscience in theory and application, and share experiences they had with interdisciplinary projects.

This is a recording of the lecture "The Virtual Brain - Improving Life through Simulation", held by Petra Ritter from the Charité Berlin on June 24th, 2019.

Special topics in computational neuroscience, incl. talks by guest speakers, review of new publications in the field and progress report on ongoing research projects

In this engineering course, we will cover all aspects of communication acoustics, which is the way sounds travels from a source, through a channel and finally to a receiver. We will look at the different system components involved in acoustic communication, including those between humans, between humans and machines, and between machines. This includes:

• speech acoustics
• hearing acoustics
• electroacoustics
• spatial sound capture and presentation
• simulation of acoustical environments
• the human auditory system
• digital audio processing methods

You will learn from top experts in the field of communication acoustics, who are all affiliated with TU9, the nine leading Universities of Technology in Germany. Together, they have pooled their expertise in order to teach a comprehensive basic understanding and indicate current research trends to you.

The lecture covers the theoretical foundations of neuroprostheses. As the underlying principle of all neuroprostheses is the electrical excitation of neurons, we will cover this topic in depth using cochlea implants as an example. In the practical computer laboratory (2SWS), which complements
the lecture (2SWS), we will program a solver for the cable equation of an active axon and implement a computer model of a cochlea implant.

Binaural hearing: binaural cues, masking, directional hearing, movement perception, precedence effect, models; Hearing impairment: Kinds of hearing impairment, frequency selectivity and auditory filters, masking and across-frequency processes, loudness and recruitment, temporal and
spectral processing, pitch perception, models of peripheral processing; Speech understanding: cues, models (Articulation Index, Speech Intelligibility Index), binaural speech understanding, effect of noise and reverberation on speech understanding; Auditory scene analysis; Music perception: Harmony,
consonance, dissonance; Hearing aids: function and algorithms; Cochlear implants: function, algorithms, temporal and spectral resolution, speech understanding.

The hippocampus and surrounding medial temporal lobe (MTL) contain cell types that represent place, time and are important for memory. Both brain regions are conserved across species implying general function. For years this function has been suggested in pattern separation, pattern completion, and attractor dynamics that would be necessary for memory. This winter term we first examine recent evidence for pattern completion, pattern separation and attractor dynamics based on the anatomy and circuitry of the hippocampus. Later we will broaden the view and look at papers on how hippocampus and MTL are related to time and space.

Introduction to computer science, computer programming, and data
processing. Differences between programming procedural vs object
oriented; discussing efficiency of code execution vs. simplicity of
programming; introduction to programming in C, python and matlab (or
similar respective environments).

For further information please contact the lecturer Jakob Macke

The goal of this practical course is to give students the tools, knowledge and hands-onexperience needed to plan, conduct and analyse a task-based fMRI or PET experiment. In the first week of the course, there will be theoretical lectures on data acquisition and analysis as well as guided tutorials on how to analyse fMRI and structural MRI data with SPM12 and Melodic in FSL. The tutorials are self-paced. In the second week, you will be asked to analyse a data set on your own and write a short report to be handed in at the end of the course.

For further information please contact the lecturer Christian Leibold.

Interdisciplinary lecture series taught by neuroscience experts from TUM and LMU that provides an introduction to computational neuroscience.
General overview: Anatomical and physiological basis of neuroscience
Modeling: Neural dynamics and coding
Towards integration in the nervous system Engineering for Neuroscience and Neuroprothetics

The lecture gives an introduction to the following topics: Neurons & glia, passive membrane properties, ion channels, cation potentials, synaptic transmission, transmitter systems, cellular networks, motor systems, learning and memory, sensory systems, orientation, echolocation.

For more information please contact the lecturers Leibold and Wachtler

This course covers the mathematical foundations of cellular physiology, ranging from the ionic basis of the membrane potential to electrochemical signaling to the propagation of action potentials in axons and dendrites of neurons based on the Hodgkin-Huxley model of the squid giant axon.
Students learn the basics of dynamical systems theory and computational neuroscience.

This intense block course provides some theoretical background,
extensive hands-on programming exercises in Matlab and interpretation of the obtained modeling results. The course is structured in 6 weeks:
Weeks 1&2: Spike train analysis, statistical models. Weeks 3&4:
Biophysical models of single neurons. Weeks 5&6 Small network models

This intense block course provides some theoretical background, extensive hands-on programming exercises in Matlab and interpretation of the obtained modeling results. Multi-channel neurophyisological data analysis: Data pre-processing, data analysis toolboxes, theory of multi-dimensional statistical methods, matlab implementation, visualization

Students perform an individual research project (lab rotation) in the Auditory Neuroscience Group. Topics and methods depend on the individual interests and backgrounds of the students. Vertebrate auditory system: Animal behavior, electrophysiology, anatomy, or data analysis project.

Students perform an individual research project (lab rotation) in the Computational Neuroscience Group. Topics and methods depend on the individual interests and backgrounds of the students. Invertebrate (leech) mechanosensory system: Electrophysiology, data analysis or modeling project. (Intracellular recordings and simulation techniques can be learned during the project.)  Vertebrate auditory system: Modeling project.

Introduction to Matlab programming for students with little or no programming experience. In hands-on exercises, students first practice to use basic programming concepts (scripts and functions, data flow and control flow, loops, structures and cell arrays…) and then apply their knowledge to statistical analysis of biological (including neuroscientific) data. After the block course, students work independently on a programming task used for evaluation.

This background module in neurophysiology consists of three weeks of seminar and hands-on lab exercises on intracellular recordings from leech neurons, as well as computer simulations to study the basis of membrane potential and action potential generation. The seminar covers the following topics:
• Invertebrate neuronal systems in comparison to vertebrate systems
• Ion channels, membrane potential and action potential generation
• Introduction to electrophysiological methods
• Introduction to data analysis methods In the practical exercises, portfolio assignments will be performed on:
• Qualitative electrophysiological classification of different cell types in the leech nervous system
• Quantitative analysis (stimulus - response relationship) of at least one cell type
• Action potential generation: Comparison of simulation and experiment

Introduction to Matlab, Readable Code: good practice, Data Import, basic statistics, frequency analysis, Continuous data (EEG/LFP), Behavioral  data, Flow control, Spike data, Logical Indexing, complex data types, Variables and their scope, Advanced plotting & Matrix operations, Images, Object-oriented programming, Individual projects, Version control, Wrap up, Principles of data analysis.

Here you find recordings of the course Theoretical Neuroscience I, held by Jochen Braun at the University of Magdeburg.

Individual topics:

1. Membrane potential and Nernst equation
2. Single compartment model
3. Leaky integrate and fire model 
4. Hodgkin-Huxley model
5. Hodgin-Huxley experiments and Connor-Stevens model (all below available soon)
6. Synaptic conductances
7. LIF nets
8. Receptive fields and tuning curves
9. Linear filter models in 2D and 3D
10. Psychophysics and psychometric functions
11. Neurometric functions
12. Population decoding with neurons
13. Shannon information basics
14. Conditional entropy and and mutual information

To go to directly to a specific lecture, right click the link (blue) and choose "open in new tab".

Here you find recordings of the course Theoretical Neuroscience II, held by Jochen Braun at the University of Magdeburg.

Individual topics:

1. Firing rate models
2. Feedforward networks
3. Assiciative memory
4. State-space analysis
5. Hebbian models of plasticity
6. Stable Hebbian plasticity
7. Hebbian learning and development
8. Reinforcement learning I - Rescorla-Wagner rule
9. Reinforcement learning II - Temporal difference learning
10. Reinforcement learning III - Policy learning
11. Reinforcement learning IV - Actor-critic models
12. Causal models
13. Principal component analysis
14. Sparse coding
15. Independent component analysis

In this lecture, and its follow-up Action & Cognition II in the summer term, we discuss the physiological substrate of cognitive processes with an emphasis on their relation to behavior. On your journey through the brain we will meet object recognition, attention, decision processes, movement planning and consciousness. A bias will be on physiological mechanisms, but due attention to clinical aspects, theoretical analysis and information theoretic measures will be given.

In this lecture, it is the follow-up of Action & Cognition I in the winter term, we discuss the physiological substrate of cognitive processes with an emphasis on their relation to behavior. On your journey through the brain we will meet object recognition, attention, decision processes, movement planning and consciousness. A bias will be on physiological mechanisms, but due attention to clinical aspects, theoretical analysis and information theoretic measures will be given.

The course focuses on the cognitive basis relevant for the design of user  interfaces, the development of user interfaces, and usability aspecs of user interfaces.

Various topics

Various topics

Bayesian statistics and modeling, JAGS

The course starts with the basics of image processing and proceeds to computer vision, A focus is on object recognition.

The course gives an introduction to unsupervised and supervised techniques of machine learning and data mining: Decision trees, clustering, dimension reduction, classification and artificial neural networks.

Important frameworks for advanced methods in Machine Learning are discussed in this course. This includes e.g. SVMs, probabilisitic methods, and reinforcement learning.

Modelling spiking neurons and network, principles of complex systems

In this lecture, we will discuss cutting edge approaches from the field of Neuroinformatics. The aim of the lecture is to get the students familiar with the concept of modelling and abstracting data, and the up to date knowledge about computational processes in the brain. After a short introduction that covers probability theory, and linear models for regression and classification, we will start a journey through the fields of graphical models and liquid computing. In the last part of the lecture we will conclude with an outlook to self-organization with the purpose to optimize information processing in complex systems like the brain. To link the knowledge acquired in this course with scientific question every second lecture a 30 min 3W session is offered. The three big W are: why should I learn this / what for can I use it / how can it be important my bachelor and master thesis . The lecture will be supplemented by a block seminar on decoding neuronal activity at the beginning of the semester break.

Probability theory, judgment and decision making, choice models, signal detection theory

This course will provide a detailed overview of the experimental methods currently used in the Neurosciences to record (as well as modulate) neuronal activity – from the local activity with single-synapse resolution to population activity at the level of brain areas.

This course will provide an introduction to important topics and algorithms in machine learning. A particular focus of this course will be on algorithms that have a clear statistical (and often Bayesian) interpretation.

In this course, students will learn about important topics and techniques in machine learning, with a particular focus on probabilistic models. The course will cover supervised learning (linear regression algorithms, linear discriminants, logistic regression, nonlinear classification algorithms) and unsupervised learning (principal component analysis including several generalizations, k-means, mixture of Gaussians, Expectation-Maximization)

The lecture introduces models of neurons of different complexity from the detailed Hodgkin-Huxley models for action potential generation via
integrate-and-fire models to simple firing rate models. Based on these
specific examples basic concepts of differential equations, linear system theory, dynamical systems theory and stochastic systems are introduced. These tools are essential for modelling neural systems and other complex systems like, for example, signaling cascades and population dynamics.
Central to the module are the exercises that match the topics from the
lecture and repeat the necessary math basics.

This course treats the basic biophysics of the signal generation and
transmission in neurons and discusses how the underlying physical and physiological phenomena can be approximated by mathematical models. Typically, such models can be characterized as nonlinear dynamical systems.